Shipment ETA & Dwell Decisions
Explore context-aware shipment ETA, calibrated uncertainty, late-arrival risk, and dwell-risk decision support.
Explore the workflow
Serverless logistics lab that predicts shipment ETA from origin, destination, departure timing, lane history, facility dwell profiles, and reliability context. ETA v1 predicts a residual correction over a transparent heuristic, then uses held-out calibration residuals for a likely arrival range, P90 ETA, and operational late-arrival probability. Training and evaluation use disclosed synthetic FTL shipment history.
- ETA Prediction (Gradient Boosting Residual v1)
- 90% Likely Arrival Range + P90 ETA
- Operational Cutoff Late-Risk Probability
- Lane-Level Benchmark Evidence
- Dwell Risk Scoring (P50/P90)